Using Geographic Information Systems (GIS) and predictive models to analyse the habitat use of bottlenose dolphins Tursiops truncatus in a Special Area of Conservation (SAC)

The EU Habitats Directive considers bottlenose dolphins Tursiops truncatus a priority species for conservation. In compliance, the outer Shannon estuary on the west coast of Ireland, has been selected as a candidate Special Area of Conservation (cSAC) to help maintain a favourable conservation status of its resident dolphin population. This paper contributes to ongoing studies in the area investigating dolphin distribution by analysing the influence of three physical variables, depth, benthic slope and aspect. The data used in this study were compiled from a 13 year dataset collected during repeated, standardised, boat based surveys covering a particular 150 km2 area of the outer estuary. Biased survey effort was controlled by calculating sightings rate for Tursiops truncatus in 1 km2 grid cells, dividing the number of sightings by number of searches. Using a Geographical Information System (GIS), distribution maps were created, enabling spatial analysis of habitat use. Relevant bathymetry and aspect data were extracted from the GIS and used in the data modelling. Generalised Additive Models (GAMs) were used to examine the influence of the predictor variables on sightings rate (under a quasipoisson distribution) and sightings occurrence (under a binomial distribution). Time in the tidal cycle notably affected dolphin distribution, with more concentrated clustering on the ebb and low water, while more spread on the flood and high water. Furthermore, the 50 % harmonic mean isopleths only overlapped by 28.5 %. The binomial model selected depth as the most significant predictor of dolphin occurrence, showing a non linear relationship, which peaked in cells with values of 16-20 m and explained 5.65 % of the variance (p < 0.001). The quasipoisson model selected aspect as the most significant predictor of sightings rate, explaining 3.95 % of the variance, with easterly facing and flat slopes the more significant (p < 0.01). Since the variables analysed in this study exerted little influence on dolphin distribution, it is likely that other more important factors, including fluid oceanographic, biological and anthropogenic variables, may better explain it. This paper has therefore successfully demonstrated how GIS and predictive modelling are powerful tools in ecology, particularly when used in combination to analyse habitat use.Faculty of Science & Technolog